Home Mathematics Two characterizations of finite nilpotent groups
Article Publicly Available

Two characterizations of finite nilpotent groups

  • EMAIL logo
Published/Copyright: August 7, 2018

Abstract

In this note we give two characterizations of finite nilpotent groups. First, we show that a finite group G is not p-nilpotent if and only if it contains two elements of order qk, for q a prime different than p, whose product has order p or possibly 4 if p=2. We also show that the set of words on two variables where the total degree of each variable is ±1 can be used to characterize finite nilpotent groups. Using this characterization we show that if a finite group is not nilpotent, then there is a word map of specified form for which the corresponding probability distribution is not uniform.

1 Introduction

In this note we give two characterizations of finite nilpotent groups. The first characterization is an observation about the order of the product of elements in a finite group which is not nilpotent. Baumslag and Wiegold [1] showed that a finite group is nilpotent if and only if the product of elements of coprime orders m and n has order mn. We show that a stronger condition holds. Recall that a finite group is said to be p-nilpotent if it has a normal p-complement. A group G is nilpotent if and only if it is p-nilpotent for every prime p dividing the order of |G|. We show the following:

Theorem A.

Let G be a finite group. Then G is not p-nilpotent if and only if there are two elements g,hG with o(g)=o(h)=qk for some prime qp and

{o(gh)=pif p is odd,o(gh){2,4}if p=2.

Our other characterization of finite nilpotent groups is slightly more technical and before presenting it, we will introduce some definitions. Let G be a finite group and let Gk be the direct product of k copies of G. Let w= be a word on d variables, i.e., w is an element of the free group F[x1,,xd]. The word w induces a map from Gd to G by substitution of the variables xi with elements of Gd. Define the function μG,w on G by

μG,w(g)=1|G|d#{(g1,,gd)Gd:w(g1,,gd)=g}.

We note that μG,w is a probability distribution on G. In recent work, Nikolov and Segal [4] have shown the following:

Theorem 1.1.

A finite group G is solvable if and only if μG,w(1) is bounded away from zero as w ranges over all words.

Work by Bray, J. S. Wilson, and R. A. Wilson [2] described a sequence of words such that the corresponding probabilities can identify when a finite group G is solvable. In this note, we do likewise for nilpotent groups, giving yet another characterization of finite nilpotent groups.

Let Totw(x) be the total degree of x in w, i.e., the total number of times that x appears in w up to multiplicity with x-1 having multiplicity -1. Similarly Totw(y) is the total degree of y in w. Let 𝒫 be the set of all words w on two variables with Totw(x)=±1 and Totw(y)=±1.

We note that a word w on two variables defines a binary operation *w on G, where g*wh=w(g,h). Write (G,*w) for the structure consisting of the underlying set of G together with the operation *w.

Recall, that a structure (G,*), where * is a binary operation, is a quasigroup if the equation x*y=z can be solved for any single unknown value uniquely. That is to say that any two entries of the tuple (x,y,x*y) determine the third. We prove:

Theorem B.

A word wF[x,y] is in P if and only if the length of w is greater than 1 and for any abelian group G with order less than or equal to the length of w, the structure (G,*w) is a quasigroup.

We will show that for any nilpotent group G and word w𝒫, the structure (G,*w) is a quasigroup. Consequently, we have:

Lemma 1.2.

If G is a nilpotent group and wP, then μG,w(g)=1|G| for all gG, i.e., μG,w is uniform.

Hence the class of nilpotent groups can be used to define 𝒫. In the finite case, the class 𝒫 can be used to characterize the nilpotent groups:

Theorem C.

A finite group G is nilpotent if and only if for all w in P with length less than 4|G|, we have μG,w(1)=1|G|.

We note that Lemma 1.2 implies the forward direction of Theorem C.

2 Groups that are not p-nilpotent

We will use the Frobenius Complement Theorem as found in Isaacs [3]:

Theorem 2.1.

Let G be a finite group, and suppose p is a prime. Then the following are equivalent:

  1. G has a normal p-complement.

  2. 𝐍G(X) has a normal p-complement for every nonidentity p-subgroup XG.

  3. 𝐍G(X)/𝐂G(X) is a p-group for every p-subgroup XG.

We will also use the following lemma, taken from an exercise in Isaacs.

Lemma 2.2 ([3, Exercise 4D.4]).

Let A act via automorphisms on G, where G is a p-group and (|G|,|A|)=1. Suppose that A acts trivially on every A-invariant proper subgroup of G, but that the action of A on G is nontrivial. Then the exponent of G is p or 4.

We now prove Theorem A restated below:

Theorem 2.3.

Let G be a finite group. Then G is not p-nilpotent if and only if there are two elements g,h in G with o(g)=o(h)=qk for some prime qp and

{o(gh)=pif p odd,o(gh){2,4}if p=2.

Proof.

Clearly the existence of two such elements implies that G is not p-nilpotent. Suppose now that G is not p-nilpotent. Since G does not have a normal p-complement for the prime p, by the Frobenius Complement Theorem, there is a p-subgroup H<G, such that |𝐍G(H):𝐂G(H)| is divisible by a prime qp. We can assume H is minimal with this property, i.e., for all p-subgroups K<H, the groups 𝐍G(K)/𝐂G(K) are p-groups. Let Q be a Sylow q-subgroup of 𝐍G(H). Then Q acts on H via automorphisms nontrivially, and by the minimality of H, we see that Q centralizes every Q-invariant subgroup of H. By Lemma 2.2 the exponent of H is either p or 4.

Let xH such that for some tQ we have xtx. Then

xtx-1=t-1xtx-1=t-1(xtx-1).

Since Q normalizes H, xtx-1H and has order either p or 4. The element t is in Q and thus o(t-1)=o(xtx-1)=qk. ∎

We will use Theorem A to prove Theorem C.

3 Words in nilpotent groups

The next lemma examines how the structure (G,*w) relates to (G/𝐙,*w).

Lemma 3.1.

Let G be a group and let Z=Z(G). Let wP. The structure (G,*w) is a quasigroup if and only if (G/Z,*w) is a quasigroup.

Proof.

Suppose that (G,*w) is a quasigroup. Then w(a,b)=w(a,c) implies b=c. Use the “overbar notation” for the correspondence between elements of G and G/Z and suppose w(a¯,b¯)=w(a¯,c¯). Then

w(a,b)¯=w(a,c)¯

and we conclude that w(a,b)=w(a,c)z for some zZ. Since w𝒫, we see that w(a,c)z=w(a,cz±1) and thus b¯=c¯. By symmetry, w(a¯,b¯)=w(c¯,b¯) implies that a¯=c¯ and (G/Z,*w) is a quasigroup.

Suppose that (G/Z,*w) is a quasigroup and w(a,b)=w(a,c). Then b¯=c¯ and we conclude that c=bz for some zZ. But

w(a,c)=w(a,bz)=w(a,b)z±1=w(a,b)

and we conclude that z=1 and thus c=b. By symmetry, we see that (G,*w) is a quasigroup. ∎

We immediately have the following corollary:

Corollary 3.2.

Let G be a nilpotent group and let wP. Then (G,*w) is a quasigroup.

We now prove Theorem B which we restate below:

Theorem 3.3.

A word wF[x,y] is in P if and only if the length of w is greater than 1 and for any abelian group G with order less than or equal to the length of w, the structure (G,*w) is a quasigroup.

Proof.

As stated in the above corollary to Lemma 3.1, for any nilpotent group G and any word w in 𝒫 the structure (G,*w) is a quasigroup.

Now suppose that w is not in 𝒫 and the length of w is greater than 1. We will demonstrate that there is a cyclic group G, with order less than or equal to the length of w, for which (G,*w) is not a quasigroup. Without loss of generality, either Totw(x)=0 or there is a prime p that divides Totw(x). If Totw(x)=0, then for any abelian group g*1=1 for all gG, and we have that (C2,*w) is not a quasigroup. Suppose Totw(x) is divisible by p. Then consider Cp, the cyclic group of order p, and note that g*1=1 for all gG. Therefore (Cp,*w) is not a quasigroup. Note that either Totw(y)=0 (and hence (C2,*w) is not a quasigroup), or p is strictly less than the length of w. ∎

Note that a word w is in 𝒫 if and only if (G,*w) is a quasigroup for all nilpotent groups G. Theorem C provides a strong converse, i.e., a group G is nilpotent if and only if (G,*w) is a quasigroup for all w in 𝒫. We now prove Theorem C which we restate below:

Theorem 3.4.

A finite group G is nilpotent if and only if for all w in P with length less than 4|G|, the function μG,w(g)=1|G| for all gG.

Proof.

For a nilpotent group G the structure (G,*w) is a quasigroup; therefore μG,w(g)=1|G| for all gG. Suppose that G is not nilpotent. We will construct a word w, so that (G,*w) is not a quasigroup. In particular, suppose that G is not p-nilpotent for the prime p. Then by Theorem A there are two elements g and h in G of order qk, for a prime q distinct from p, such that o(gh)=p or 4. We will abuse notation and let p=o(gh). Let 0<a<qk be an inverse of p modulo qk, that is ap1(modqk). Since (a,qk)=1, there are unique g^ and h^ such that g^a=g and h^a=h. Moreover, o(g^)=o(h^)=qk. Let b satisfy ap-bqk=1.

If o(gh)=p, consider the word

w(x,y)=(xaya)p-1(xa-bqkya-bqk).

We note the following observations about w(x,y):

  1. Totw(x)=Totw(y)=ap-bqk=1. Hence w(g,g-1)=1 for all gG.

  2. w(g^,h^)=(gh)p-1(g^a-bqkh^a-bqk)=(gh)p=1.

Since g^ is not h^-1 (since gh-1), we conclude that μG,w(1)>1|G|.

The length of w is 2(ap+bqk). We note that a<qk and bqk=ap-1<|G|. Hence the length of w is less than 4|G|. ∎

We remark that analogs of Theorem B and Theorem C are still true if 𝒫 is replaced with the class of all words w with Totw(x)=Totw(y)=1. Such words will generate a loop on every nilpotent group.


Communicated by Evgenii I. Khukhro


Funding statement: The author acknowledges that this material is based upon work supported by the National Science Foundation Graduate Research Fellowship Program under Grant No. DGE-1256529.

Acknowledgements

References

[1] B. Baumslag and J. Wiegold, A sufficient condition for nilpotency in a finite group, preprint (2014), https://arxiv.org/abs/1411.2877. Search in Google Scholar

[2] J. N. Bray, J. S. Wilson and R. A. Wilson, A characterization of finite soluble groups by laws in two variables, Bull. Lond. Math. Soc. 37 (2005), no. 2, 179–186. 10.1112/S0024609304003959Search in Google Scholar

[3] I. M. Isaacs, Finite Group Theory, Grad. Stud. Math. 92, American Mathematical Society, Providence, 2008. Search in Google Scholar

[4] N. Nikolov and D. Segal, A characterization of finite soluble groups, Bull. Lond. Math. Soc. 39 (2007), no. 2, 209–213. 10.1112/blms/bdl028Search in Google Scholar

Received: 2017-05-31
Revised: 2018-03-14
Published Online: 2018-08-07
Published in Print: 2018-11-01

© 2018 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 25.3.2026 from https://www.degruyterbrill.com/document/doi/10.1515/jgth-2018-0029/html
Scroll to top button